Hepatocellular carcinoma is among the most common malignant tumors worldwide, ranking sixth in incidence and third in mortality. Surgical resection offers the best chance of cure for early-stage disease, but because the cancer often develops silently, most patients are diagnosed at intermediate or advanced stages when surgery is no longer an option. For these patients, treatment has been transformed in recent years by the arrival of immune checkpoint inhibitors targeting the PD-1 protein and agents that block vascular endothelial growth factor (VEGF), a driver of tumor blood supply. Bevacizumab, a VEGF inhibitor, suppresses tumor neovascularization and reshapes the immune environment around the tumor, while sintilimab, a fully human IgG4 monoclonal antibody against PD-1, enhances the immune system’s capacity to attack tumor cells. The combination of a bevacizumab biosimilar with sintilimab has been approved in China as first-line therapy for unresectable disease, and a multicenter phase 2–3 study found it significantly outperformed sorafenib, cutting the risk of death by 43 percent and the risk of progression by 44 percent.

The triple therapy examined in the new study adds transarterial chemoembolization, or TACE, to the bevacizumab-sintilimab duo. In TACE, an interventional radiologist threads a catheter to the artery feeding the tumor and injects a mixture of iodized oil and chemotherapy drugs, in this protocol oxaliplatin and pirarubicin, followed by gelatin sponge particles that block the vessel. This delivers high drug concentrations directly to the tumor while cutting off its blood supply. Beyond its local effect, TACE appears to shift the immunosuppressive tumor microenvironment toward one that supports immune attack, potentially amplifying the benefit of PD-1 blockade. Recent studies of the triple approach in advanced hepatocellular carcinoma have reported objective response rates between roughly 63 and 71 percent, median progression-free survival of 6.0 to 12.4 months, and median overall survival of 12.2 to 22.3 months, outperforming doublet therapy and traditional TACE alone with manageable side effects.

Despite these encouraging results, clinicians have lacked a validated tool to forecast which patients will benefit from the triple regimen. The research team, led by investigators at Peking University People’s Hospital and collaborating institutions, recruited 156 patients with unresectable hepatocellular carcinoma who received TACE combined with bevacizumab and sintilimab as first-line treatment between January 2021 and May 2024. Patients came from Peking University People’s Hospital and Nanfang Hospital, which formed an 86-patient training cohort, and from the First Affiliated Hospital of Guilin Medical University, which provided an independent external validation cohort of 70 patients. All participants had an ECOG performance status of 0 or 1, at least one measurable tumor by modified RECIST criteria, and Child-Pugh liver function of class A or B.

The treatment outcomes in the combined cohort were substantial. Forty-three patients, or 27.6 percent, achieved a complete radiological response, and 53, or 34.0 percent, achieved a partial response, yielding an objective response rate of 61.5 percent and a disease control rate of 82.1 percent. Notably, 23 patients responded well enough to meet surgical criteria and underwent sequential resection, suggesting the regimen may serve as a conversion strategy in selected cases. Treatment-related adverse events were common but controllable: about 82.6 percent of the training cohort experienced some adverse event, most frequently abnormal liver function tests, fever, and nausea, and most moderate to severe events were managed with dose adjustment or temporary drug withdrawal. No treatment-related deaths occurred.

To build the predictive model, the researchers first used receiver operating characteristic analysis to determine the optimal cutoff for the alkaline phosphatase to prealbumin ratio, finding a threshold of 0.96. Univariate Cox regression in the training cohort identified elevated total bilirubin above 17 micromoles per liter, alpha-fetoprotein above 100 nanograms per milliliter, an APR above 0.96, and BCLC stage C as adverse prognostic factors for overall survival. Multivariate analysis then narrowed the field to three independent predictors: APR above 0.96 carried a hazard ratio of 3.360, BCLC stage C a hazard ratio of 3.075, and elevated bilirubin a hazard ratio of 1.896. Alpha-fetoprotein dropped out of the final model. Weighting each variable by its regression coefficient produced the ABT score, with patients classified as low risk at or below 1.0, intermediate risk between 1.0 and 2.0, and high risk above 2.0.

The model’s discriminatory power was strong and consistent across cohorts. The area under the ROC curve for predicting overall survival was 0.823 in the training cohort and 0.828 in the validation cohort, exceeding the performance of any single component variable, whose AUC values ranged from 0.625 to 0.733. The C-index, a measure of concordance between predicted and observed survival rankings, was 0.792 and 0.795 in the two cohorts respectively. The ABT score also outperformed traditional liver function assessments: in the training cohort, ALBI scored an AUC of 0.681 and Child-Pugh 0.662, compared with 0.823 for ABT, and similar gaps held in validation. The score remained stable over time, with one-year and two-year AUC values holding in the 0.82 to 0.87 range in both cohorts, and calibration curves showed good agreement between predicted and observed survival at both time points.

The risk stratification the model produced was striking. In the training cohort, two-year overall survival was 89.8 percent for the low-risk group, 43.7 percent for the intermediate-risk group, and 12.9 percent for the high-risk group, a difference that was highly statistically significant. Two-year progression-free survival followed the same gradient at 87.6, 76.7, and 30.8 percent. In the validation cohort, two-year overall survival was 72.1, 24.6, and 9.6 percent across the three risk tiers. Because patients who undergo conversion surgery may artificially inflate survival estimates through immortal time bias, the researchers repeated the analyses excluding these patients, and the ABT score retained its stable ability to separate the remaining 133 patients into three distinct survival groups.

The biological logic behind the score’s components is grounded in tumor and liver physiology. Elevated alkaline phosphatase may reflect bile stasis injury, hepatic parenchymal stress, biliary or vascular damage, and a higher intrahepatic tumor burden, while falling prealbumin signals impaired hepatic synthetic capacity, malnutrition, systemic inflammation, and reduced physiological reserve. A rising APR therefore points to a convergence of aggressive tumor biology and weakening liver function, a combination particularly consequential in triple therapy, where chemoembolization stresses the liver and both anti-VEGF treatment and immune checkpoint inhibition require adequate hepatic reserve. BCLC stage contributes standardized tumor staging, and bilirubin reflects the degree of hepatocyte injury.

The authors emphasize several limitations. The retrospective design and modest sample size introduce potential selection bias, and TACE procedures were not fully standardized across the three centers. Follow-up was relatively short, and median overall survival had not yet been reached in the validation cohort. Furthermore, 83.3 percent of patients had hepatitis B infection, so the findings require confirmation in populations with other causes of liver cancer, ideally in large prospective multicenter studies. The researchers also note that the ABT score should be viewed as a complement to, not a replacement for, established tools such as ALBI and Child-Pugh, since it adds tumor stage and the nutritional-inflammatory dimension to purely hepatic measures. If validated further, the score could offer clinicians a simple, noninvasive way to identify which patients with unresectable hepatocellular carcinoma are most likely to benefit from this triple therapy, and to guide treatment decisions at the bedside using nothing more than routine blood work and standard staging.

Subject of Research: A prognostic scoring model for unresectable hepatocellular carcinoma patients receiving TACE combined with bevacizumab and sintilimab therapy

Article Title: ABT score based clinical model predicts prognosis in unresectable hepatocellular carcinoma treated with TACE combined with bevacizumab and sintilimab therapy

Article References: Liao, M., Wang, X., Wang, Z., Ai, X., Chen, Z., Wang, X., Wang, K., Fei, X., Qin, W., Liu, F., & Rao, H. (2026). ABT score based clinical model predicts prognosis in unresectable hepatocellular carcinoma treated with TACE combined with bevacizumab and sintilimab therapy. Frontiers in Pharmacology, 17, Article 1756451. https://doi.org/10.3389/fphar.2026.1756451

Image Credits: AI Generated

DOI: 10.3389/fphar.2026.1756451

Keywords: hepatocellular carcinoma, ABT score, TACE, bevacizumab, sintilimab, prognosis prediction, overall survival, biomarker, immune checkpoint inhibitor, liver cancer, risk stratification, triple therapy